人工神经网络模型在肺癌与胃癌或肠癌中的鉴别分析  

The distinguishment of lung cancer with gastric cancer or colon cancer by artificial neural network

作　　者：周晓蕾[1] 冯斐斐[2] 张昭[3] 秦利娟[2] 吴拥军[2] 聂广金[2] 倪然[1] 吴逸明[2] 王静[1] 

机构地区：[1]郑州大学第一附属医院呼吸内科,450052 [2]郑州大学公共卫生学院,450001 [3]郑州大学临床医学系,450001

出　　处：《实用医学杂志》2011年第18期3312-3314,共3页The Journal of Practical Medicine

基　　金：国家自然科学基金资助项目(编号:30972457)

摘　　要：目的:应用人工神经网络技术,联合检测6种肿瘤标志对肺癌与胃癌或肠癌进行区分判别,建立肿瘤标志联合检测肺癌的辅助诊断模型。方法:采用放射免疫学、分光光度法、原子吸收分光光度法等方法,测定67例肺癌患者、47例胃癌患者和50例大肠癌患者血清中癌胚抗原(CEA)、胃泌素(gastrin)、神经元特异性烯醇化酶(NSE)、唾液酸(SA)、铜锌比值(Cu/Zn)、钙(Ca)等6项指标。建立基于人工神经网络的肺癌肿瘤标志智能诊断模型。结果:肺癌-胃癌的人工神经网络模型判别肺癌的灵敏度,特异度和准确度分别为100%、83.3%和93.5%;肺癌-肠癌模型判别肺癌的灵敏度、特异度和准确度分别为76.9%、100%和87.0%。结论:本研究成功建立基于人工神经网络技术的肿瘤标志物联合检测的人工智能诊断模型,对肺癌-胃癌、肺癌-肠癌中肺癌的鉴别诊断有助于提高肺癌的诊断率。Objective To distinguish lung cancer from gastric cancer or colon cancer by artificial neural network （ANN） combined with six serum tumor markers. Methods The levels of serum carcino-embryonic antigen （CEA）, gastrin, neurone specific enolase （NSE）, sialie acid （SA）, Cu/Zn, Ca in 67 lung cancer patients, 47 gastric cancer patients and 50 colon cancer were detected by radioimmunology,spectrophotometry,or atomic absorption spectrophotometry,respectively.and artificial neural network were established with six serum tumor markers to distinguish lung cancer from the other cancers. Results The sensitivity, specificity and accuracy of distinguishing lung cancer by lung cancer-gastric cancer ANN model were 100%, 83.3% and 93.5%, respectively. And by lung cancer-colon cancer were 76.9%, 100% and 87.0%. Conclusions There is a clinical significant effect to distinguish lung cancer from gastric cancer and colon cancer by ANN model combined with optimal serum markers, which is very helpful for diagnosis of lung cancer.

关 键 词：神经网络(计算机) 肺癌 肿瘤标志 胃癌 肠癌 

分 类 号：R734.2[医药卫生—肿瘤] R735.2[医药卫生—临床医学]